import pandas as pd
import numpy as np
import pickle
import json

# Visualisation libraries

# Maps
import folium

## Text
from colorama import Fore, Back, Style
from IPython.display import Image, display, Markdown, Latex, clear_output

## seaborn
import seaborn as sns

## matplotlib
import matplotlib.pyplot as plt
from matplotlib.patches import Ellipse, Polygon
from matplotlib.font_manager import FontProperties
import matplotlib.colors as mcolors
plt.style.use('seaborn-whitegrid')
plt.rcParams['axes.labelsize'] = 14
plt.rcParams['xtick.labelsize'] = 12
plt.rcParams['ytick.labelsize'] = 12
plt.rcParams['text.color'] = 'k'
%matplotlib inline

import warnings
warnings.filterwarnings("ignore")

Home / Healthcare Analytics and Modeling

Los Angeles County Restaurant and Market Violations and Inspections

• Preprocessing
• Exploratory Data Analysis
• Interactive Maps

Los Angeles County Restaurant and Market Violations and Inspections

In this article, we use a dataset contains Environmental Health Violations for Restaurants and Markets in Los Angeles County. This dataset can be found [1] and [2].

Los Angeles County Environmental Health is responsible for checking food violations for all unincorporated areas and 85 of the 88 cities in the County. This dataset does not include Pasadena, Long Beach or Vernon (each has its own city health department).

Loading Datasets

Violations_df = pd.read_csv('LA/Clean_Violations_df.csv')
Inspections_df = pd.read_csv('LA/Clean_Inspections_df.csv')
Data  = pd.read_csv('LA/Data_Inspections_Violations_df.csv')
Violations_matrix = pd.read_csv('LA/Violations_matrix_df.csv')
# Dictionaries
with open('LA/vDict.pkl', 'rb') as fp:
    Violation_Dictionary= pickle.load(fp)

Data['Facility Zip'] = Data['Facility Zip'].astype(str)
Data['Facility Zip'] = Data['Facility Zip'].apply(lambda x: x[:5])

Geographical Analysis

Facilities_Geographical_df = Data.groupby('Facility Zip').agg(Average_Score=('Score', 'mean'),
                                                              Total_Facilities=('Facility ID', pd.Series.nunique)).reset_index()
Facilities_Geographical_df = Facilities_Geographical_df.rename(columns = {'Average_Score':'Average Score',
                                                                          'Total_Facilities':'Total Facilities'})
Facilities_Geographical_df = Facilities_Geographical_df.dropna()
Facilities_Geographical_df.head().style.hide_index() 

Facility Zip	Average Score	Total Facilities
90001	90.785648	269
90002	92.230228	70
90003	91.138814	257
90004	92.719656	258
90005	91.042819	272

Next, we can export a geojson file from here. However, we would like to reduce the size of this file for the sake of having an optimal computation.

# loading the GeoJSON file
with open('LA/LA_ZIP_Codes.geojson', 'r') as jsonFile:
    Zipcode_Data = json.load(jsonFile)

# Creating a list from Zip codes
mylist = Facilities_Geographical_df['Facility Zip'].unique().tolist()

# removing ZIP codes that are not in our dataset
temp = []
for i in range(len(Zipcode_Data['features'])):
    if Zipcode_Data['features'][i]['properties']['name'] in mylist:
        temp.append(Zipcode_Data['features'][i])

# creating a new JSON file
Reduced_Zipcode_Data = dict.fromkeys(['type','features'])
Reduced_Zipcode_Data['type'] = 'FeatureCollection'
Reduced_Zipcode_Data['features'] = temp

del Zipcode_Data, mylist, temp

# save the JSON file
open("LA/reduced_LA_ZIP_Codes.json", "w").write(json.dumps(Reduced_Zipcode_Data,
                                                             sort_keys=True, indent=4, separators=(',', ': ')))
del Reduced_Zipcode_Data

def plot_map(Inp_Column, Text_Legend = '', Inp_Df = Facilities_Geographical_df, Zoom_Level=8):

    # reading of the updated GeoJSON file
    Geographical_Data = r'LA/reduced_LA_ZIP_Codes.json'

    # initiating a Folium map
    m = folium.Map(location = [34.052, -118.243], zoom_start = Zoom_Level)

    # creating a map
    m.choropleth(geo_data = Geographical_Data, fill_opacity = 0.8, line_opacity = 0.2,
        data = Inp_Df, key_on = 'feature.properties.name', columns = ['Facility Zip', Inp_Column],
        fill_color = 'RdYlGn', legend_name = Text_Legend)
    folium.LayerControl().add_to(m)
    
    # Show the map
    return m

plot_map('Average Score', 'The Average Score of Facilities')

Make this Notebook Trusted to load map: File -> Trust Notebook

plot_map('Total Facilities','Total Facilities')

Make this Notebook Trusted to load map: File -> Trust Notebook

Code Violations by Each Facility

Frist, let's addd the Violation Matrix to our Data

Data= pd.merge(Data, Violations_matrix, on='Facility ID', how='right')

Moreover, previously, we found the following violation code as the most correlated ones.

F035, W017, CV20, CV29, W021, W011, F033, F040, F044

Thus,

Most_Corr = ['F035', 'W017', 'CV20', 'CV29', 'W021', 'W011', 'F033', 'F040', 'F044']

Facility_Violations_df = Data.groupby(['Facility Zip','Facility ID'])[Most_Corr].agg({np.mean})
Facility_Violations_df = Facility_Violations_df.groupby(level=0).mean()
Facility_Violations_df.reset_index(inplace=True)
Facility_Violations_df.columns = Facility_Violations_df.columns.droplevel(1)
display(Facility_Violations_df.head().style.hide_index())

Facility Zip	F035	W017	CV20	CV29	W021	W011	F033	F040	F044
90001	1.602230	0.000000	0.000000	0.000000	0.000000	0.000000	1.750929	0.940520	2.167286
90002	1.485714	0.000000	0.000000	0.000000	0.000000	0.000000	1.714286	0.800000	1.800000
90003	2.007782	0.000000	0.000000	0.000000	0.000000	0.000000	2.400778	1.077821	2.482490
90004	1.554264	0.000000	0.000000	0.000000	0.000000	0.000000	1.414729	0.872093	1.686047
90005	1.511029	0.000000	0.000000	0.000000	0.000000	0.000000	2.246324	1.238971	2.352941

temp = Facility_Violations_df[0:1].any().to_dict()
mylist=list({k:v for k, v in temp.items() if v == True}.keys())

for vCode in mylist[1:]:
    display(plot_map(vCode, 'Average Violation: %s' % Violation_Dictionary[vCode] ,Facility_Violations_df))

Make this Notebook Trusted to load map: File -> Trust Notebook

Make this Notebook Trusted to load map: File -> Trust Notebook

Make this Notebook Trusted to load map: File -> Trust Notebook

Make this Notebook Trusted to load map: File -> Trust Notebook

Program Elements: Seats

PE_Seats_Counts = pd.DataFrame(Data.groupby(['Facility Zip','PE Seats']).size().unstack(1)).reset_index()
PE_Seats_Counts.head().style.hide_index()

Facility Zip	0-10 SQ. FT.	0-30 SQ. FT.	0-999 SQ. FT.	1-1999 SQ. FT.	1-4999 SQ. FT.	1000-1999 SQ. FT.	10000+ SQ. FT.	151+ SQ. FT.	2000+ SQ. FT.	2000-4999 SQ. FT.	2000-5999 SQ. FT.	31-60 SQ. FT.	6000+ SQ. FT.	61-150	ANNUAL	Other
90001	nan	2432.000000	nan	1552.000000	nan	nan	nan	121.000000	449.000000	nan	nan	792.000000	nan	33.000000	nan	nan
90002	nan	469.000000	nan	442.000000	nan	nan	nan	18.000000	116.000000	nan	nan	73.000000	nan	20.000000	nan	nan
90003	nan	2531.000000	nan	2268.000000	nan	nan	nan	nan	200.000000	nan	nan	397.000000	nan	151.000000	nan	nan
90004	nan	1532.000000	nan	404.000000	nan	nan	nan	67.000000	343.000000	nan	nan	1225.000000	nan	831.000000	nan	14.000000
90005	nan	1859.000000	nan	524.000000	nan	nan	nan	389.000000	175.000000	nan	nan	1847.000000	nan	1278.000000	nan	nan

temp = PE_Seats_Counts[0:1].any().to_dict()
mylist = list({k:v for k, v in temp.items() if v == True}.keys())

for s in mylist[1:]:
    display(plot_map(s, 'Facility Total Count (%s)' % s ,PE_Seats_Counts))

Make this Notebook Trusted to load map: File -> Trust Notebook

Make this Notebook Trusted to load map: File -> Trust Notebook

Make this Notebook Trusted to load map: File -> Trust Notebook

Make this Notebook Trusted to load map: File -> Trust Notebook

Make this Notebook Trusted to load map: File -> Trust Notebook

Make this Notebook Trusted to load map: File -> Trust Notebook

Program Elements: Type

PE_type_counts = pd.DataFrame(Data.groupby(['Facility Zip','PE Type']).size().unstack(1)).reset_index()
PE_type_counts.head().style.hide_index()

Facility Zip	Caterer	Food Mkt Retail	Food Processing Wholesale	Food Vehicle Commissary	Food Warehouse	Interim Housing FF	LIC HTH Care Food FAC	Private School Cafeteria	Restaurant	Skc Tenant Retail Food Operator	Wholesale Food Complex
90001	nan	2001.000000	nan	nan	nan	nan	nan	nan	3378.000000	nan	nan
90002	nan	558.000000	nan	nan	nan	nan	nan	nan	580.000000	nan	nan
90003	nan	2468.000000	nan	nan	nan	nan	nan	nan	3079.000000	nan	nan
90004	nan	740.000000	nan	nan	nan	7.000000	nan	14.000000	3655.000000	nan	nan
90005	nan	699.000000	nan	nan	nan	nan	nan	nan	5373.000000	nan	nan

temp = PE_type_counts[0:1].any().to_dict()
mylist=list({k:v for k, v in temp.items() if v == True}.keys())

for s in mylist[1:]:
    display(plot_map(s, 'PE Type: %s' % s , PE_type_counts))

Make this Notebook Trusted to load map: File -> Trust Notebook

Make this Notebook Trusted to load map: File -> Trust Notebook

Program Elements: Risk

PE_Risk_Counts = pd.DataFrame(Data.groupby(['Facility Zip','PE Risk']).size().unstack(1)).reset_index()
PE_Risk_Counts.head().style.hide_index()

Facility Zip	High Risk	Low Risk	Moderate Risk	None	Seats High Risk	Seats Low Risk	Seats Moderate Risk
90001	896.000000	1011.000000	94.000000	nan	2499.000000	84.000000	795.000000
90002	243.000000	237.000000	78.000000	nan	338.000000	10.000000	232.000000
90003	667.000000	1651.000000	150.000000	nan	2056.000000	33.000000	990.000000
90004	350.000000	358.000000	32.000000	21.000000	2835.000000	63.000000	757.000000
90005	304.000000	366.000000	29.000000	nan	4465.000000	95.000000	813.000000

plot_map('Low Risk', 'PE Risk: Low Risk' ,PE_Risk_Counts)

Make this Notebook Trusted to load map: File -> Trust Notebook

plot_map('Moderate Risk', 'PE Risk: Moderate Risk' ,PE_Risk_Counts)

Make this Notebook Trusted to load map: File -> Trust Notebook

plot_map('High Risk', 'PE Risk: High Risk' ,PE_Risk_Counts)

Make this Notebook Trusted to load map: File -> Trust Notebook

---

References

1. LAC Open Data: County of Los Angeles Restaurant and Market Violations
2. LAC Open Data: County of Los Angeles Restaurant and Market Inspections

---